renatadaou/school-cafeteria-charlene-renata

Religious Private Schools Top the Ranking of Cafeteria Hygiene Violations - Methodology Post

By Charlene Lin & Renata Daou

In this post, we will explain how we have used the data: How did we find the story? For the information on the number of schools and violations, we explored the DOHMH School Cafeteria Inspections (2020-Present) dataset. Initially, we began by examining the school cafeteria dataset to identify schools with the highest number of violations and their geographical distribution. Since we had the longitude and latitude data available for each school, we easily mapped them using Datawrapper, creating a preliminary visual representation for our investigation, which we then refined and added to the main article.

Upon further analysis, we observed a significant concentration of schools in two particular neighborhoods in Brooklyn: Williamsburg and Borough Park. This observation prompted us to explore common characteristics among these schools, such as their public or private status, religious affiliations, and other relevant factors. This exploration formed the basis of our story.

Percentages and calculations

To make calculations easier, we created multiple pivot tables where we could filter information such as the location of the schools, the number and type of violations, and others.

The tables can be seen in the link here: NYC School Cafeteria Violation Sep 2021-April 2024

To help in understanding the article, we have provided a breakdown of all the percentages and calculations performed to determine the percentage of violations per school, type of school, and type of violations.

1. Percentage of Schools with Violations:

   • Out of 2059 schools, 799 have zero violations.
   • So, the number of schools with violations is 1260 schools.
   • Percentage of schools with violations = (1260 / 2059) * 100% ≈ 61.17%

2. Percentage of Religious Schools among Schools with Violations:

   • Out of 32 schools with 6 or more violations, 23 are religious.
   • Percentage of religious schools among schools with violations = (23 / 32) * 100% ≈ 71.88%

3. Percentage of Jewish Schools among Religious Schools with Violations:

   • Out of 23 religious schools with violations, 19 are Jewish.
   • Percentage of Jewish schools among religious schools with violations = (19 / 23) * 100% ≈ 82.61%

4. Percentage of Private Schools with Critical Violations: Out of 680 private schools, 328 had critical violations. Percentage of private schools with critical violations = (328 / 680) * 100% ≈ 48.24%

5. Percentage of Public Schools with Critical Violations: Out of 1379 public schools, 398 had critical violations. Percentage of public schools with critical violations = (398 / 1379) * 100% ≈ 28.85%

6. Schools with General Violations: For private schools: Total number of general violations = 777 Total number of private schools = 680 Average number of general violations per private school = 777 / 680 ≈ 1.14 For public schools: Total number of general violations = 990 Total number of public schools = 1,379 Average number of general violations per public school = 990 / 1,379 ≈ 0.72 Multivariable regression analysis

Statistical analyses reveal a significant gap between private and public schools, with private schools consistently showing more food violations in school cafeterias. This divide is particularly stark in Jewish schools. The information on which religious denomination the schools belong to was obtained from the Private School Universe Survey (PSS). Multivariable regression analysis indicates that Jewish schools tend to have higher counts of food violations compared to other schools. This suggests an association between being a Jewish school and having more food violations. Interestingly, factors such as student population and student-to-teacher ratio do not consistently affect food violations. This implies that school size and teacher-student ratio may not play a significant role in determining food violations.

Regression analysis is a methodical way of exploring the relationships between different factors to understand why something happens. In this case, we're investigating why there are issues with food in school cafeterias. We're looking at various factors that might be linked to these problems.

First, we consider the financial background of students' families. Then, we examine how students are performing academically. Are they achieving high scores on tests? We also evaluate the resources available in schools, such as whether there are enough teachers to supervise students properly. Additionally, we analyze the type of school – whether it's public or private, religiously affiliated or secular. We also look at the diversity of the student body. Finally, we consider the overall size of the student population.

Here’s the original analysis breakdown:

Y ~ X1 + X2 + X3 + X4 + X5 + X6 Y = Food violation counts (from 9/17/2021, when public schools re-opened since COVID) X1 = Student socioeconomic status (family income, poverty rate, eligible for free lunch programs or not) X2 = Educational attainment (SAT/ACT score, college admission rate) X3 = Resources (student-teacher ratio) X4 = School Type (Public/Private, Religious or not, Jewish/Catholic) X5 = Minority (percentage of students being minorities) X6 = Student body size (hypothesis is that if there are more people to cook for, the kitchen can be more messy)

Unfortunately, the socioeconomic status was not successfully obtained, and the data on minority students was only available for private schools. Since it is clear from preliminary data analysis that Jewish schools (usually 100% white) are where violations are concentrated, we decided to focus on schools being Jewish or not as a variable.

The Welch Two Sample t-test suggests significant differences in food violation counts between schools categorized by these variables. In the Multivariable Regression Analysis, the presence of Jewish schools significantly increases the expected food violation count, independent of other factors. Private schools tend to have slightly higher food violation counts than public schools. Student body size and Student-Teacher Ratio don’t show a significant effect on food violations, which overthrew the original hypothesis we had about how kitchen sanitation can be affected by the number of students to feed and the proxy of school resources.

In Final Private School Merged Dataset, we obtained standardized test scores of 192 NYC private schools and did a linear regression to conclude the relationship between students’ basic English and math proficiency levels and the school violation counts was significant. We aggregated the passing rate and took an average for each school that adopted the standardized tests. However, the tests were not adopted by private schools, with violations of as many as 12. Data points spanning the complete range of violation counts were collected and qualified as random samples selected from a bigger dataset. The result yields a p-value smaller than 0.05.

When it comes to test results, we explored the dataset from this article here. To match the schools in the DOHMH dataset, and the test results dataset, we used a method in Python called fuzzy match. In Python, the fuzzy-wuzzy library allowed us to perform fuzzy matching between strings, and link the schools with their cafeteria violations as well as their test results.

To understand the relationship between test results, the religious affiliation of the schools, and food violations in cafeterias, we made a new regression.

The analysis shows that Jewish schools, on average, have more violations compared to other schools. The math shows that as students perform better, the number of food violations slightly decreases. For every increase in proficiency, violations go down by about 0.016. This might seem small, but it’s statistically significant, meaning it's a real effect, not just random noise. The model explains only about 2.67% of the variation in rule violations between different schools. This is a pretty low number, which suggests that student performance alone doesn't tell us much about why some schools have more violations than others.

Find the code for the story here: https://github.com/renatadaou/school-cafeteria-charlene-renata https://github.com/charlenelin0824/connect-to-census